Computational and Mathematical Methods in Medicine

Volume 2015, Article ID 236896, 9 pages

http://dx.doi.org/10.1155/2015/236896

## Evaluation of the Acceleration and Deceleration Phase-Rectified Slope to Detect and Improve IUGR Clinical Management

^{1}Department of Obstetrical-Gynaecological and Urological Science and Reproductive Medicine, Federico II University, 5 Pansini Street, 80131 Naples, Italy^{2}Computational Physiology and Clinical Inference Group, Research Laboratory of Electronics, Massachusetts Institute of Technology, 25 Carleton Street, Cambridge, MA 02139, USA^{3}Department of Electrical, Computer and Biomedical Engineering, University of Pavia, 5 Ferrata Street, 27100 Pavia, Italy^{4}Dipartimento di Elettronica, Informazione e Bioingegneria (DEIB), Politecnico di Milano, Piazza Leonardo da Vinci 32, 20133 Milano, Italy

Received 6 September 2015; Revised 16 November 2015; Accepted 17 November 2015

Academic Editor: Joao Cardoso

Copyright © 2015 Salvatore Tagliaferri et al. This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.

#### Abstract

*Objective.* This study used a new method called Acceleration (or Deceleration) Phase-Rectified Slope, APRS (or DPRS) to analyze computerized Cardiotocographic (cCTG) traces in intrauterine growth restriction (IUGR), in order to calculate acceleration- and deceleration-related fluctuations of the fetal heart rate, and to enhance the prediction of neonatal outcome.* Method.* Cardiotocograms from a population of 59 healthy and 61 IUGR fetuses from the 30th gestation week matched for gestational age were included. APRS and DPRS analysis was compared to the standard linear and nonlinear cCTG parameters. Statistical analysis was performed through the -test, ANOVA test, Pearson correlation test and receiver operator characteristic (ROC) curves ().* Results.* APRS and DPRS showed high performance to discriminate between Healthy and IUGR fetuses, according to gestational week. A linear correlation with the fetal pH at birth was found in IUGR. The area under the ROC curve was 0.865 for APRS and 0.900 for DPRS before the 34th gestation week.* Conclusions.* APRS and DPRS could be useful in the identification and management of IUGR fetuses and in the prediction of the neonatal outcome, especially before the 34th week of gestation.

#### 1. Introduction

Intrauterine growth restriction (IUGR) is defined as a pathologic condition for a fetus that has not attained its biologically determined growth potential, for that particular gestational age. IUGR is estimated to be approximately 5–8% in the general obstetric population; frequently the etiology is the placental dysfunction [1]. It is related to an increased risk of perinatal complications, such as fetal hypoxia and asphyxia, and important long-term implications for the infant neurodevelopment. Therefore, the best time to deliver an IUGR fetus remains the most important challenge in perinatal management [2–4].

The electronic fetal heart rate (FHR) monitoring is one of the most widespread noninvasive methods to evaluate the fetal well-being during the antenatal period, especially in high risk pregnancies.

Many efforts have been made to understand the mechanisms of normal regulation of FHR variability and several studies have found that they are mainly nonlinear. Computerized Cardiotocography (cCTG) provide a standardized method to evaluate quantitative measures of linear and nonlinear indices of FHR variability [5, 6].

We used a cCTG analysis method based on a signal-processing algorithm, termed Phase-Rectified Signal Average (PRSA), that overcomes the limitations of nonstationary signal and background noise typical for FHR signal [7].

Our aim was to evaluate the trend of cCTG parameters in Healthy and IUGR fetuses, in order to detect early signs of fetal compromise and to enhance the prediction of neonatal outcome.

#### 2. Materials and Methods

This retrospective transversal study was carried out at the Department of Obstetrical-Gynaecological and Urological Science and Reproductive Medicine of the Federico II University (Italy), in collaboration with the Politecnico di Milano (Italy).

The study was conducted on a homogeneous population of 120 pregnant women composed of 59 Healthy and 61 IUGR fetuses. It was approved by the ethics committee of the university and all participants gave their written informed consent.

Inclusion criteria were Caucasian ethnicity; singleton pregnancy; certain pregnancy dating (calculated from the first day of the last menstrual period and confirmed by ultrasound measurements, according to the population nomograms) [8]; gestational age from the 30th week; and cCTGs with a signal loss of less than 15% over the whole record. We considered only the last cCTG record within 24 h of delivery and the delivery indication was only for fetal condition in IUGR group. Healthy fetuses were subjected to cCTG monitoring at the same gestational weeks of IUGR ones, but they delivered all after 37 weeks of gestation. Newborn baby data (sex, weight, Apgar score, malformation at birth, access to neonatal intensive care, and umbilical artery pH) were collected.

We excluded preexisting maternal disease, drug abuse, fetus with chromosomal and major congenital anomalies, and inadequate umbilical cord samples at birth. The severity of the growth restriction was assessed by ultrasound biometry, Doppler velocimetry of umbilical artery (UA), middle cerebral artery (MCA), ductus venosus (DV), and cCTG.

Pulsatility Index (PI) of UA and DV was considered abnormal when it was >95th centile for gestational age [9] and when absent or reverse A-wave or end-diastolic flow in DV [2, 10] and in UA was detected or MCA PI was <5th centile [11, 12].

The growth-restricted group was defined by estimated weight below the 10th centile [1] and estimated abdominal circumference below the 10th centile with abnormal UA Doppler pulsatility index (PI) > 95th centile irrespective of the presence of absent or reversed end-diastolic flow for its gestational age.

The tests were made with the same frequency in all cases.

Among 30 + 0 to 33 + 6 weeks of gestation elective caesarean section was performed in case of absent end-diastolic flow in the UA or DV PI > 95th centile with cCTG abnormalities (e.g., low short-term variation or recurrent late deceleration). After 34 + 0 weeks of gestation elective caesarean section was performed in case of PI > 95th centile in the UA or PI < 5th centile in the MCA with cCTG abnormalities (e.g., low short-term variation) [4, 13].

In IUGR group, the delivery occurred within 24 h after the administration of maternal steroids before 34 weeks. The artery umbilical gas analysis was performed after birth for all newborns [14].

To discriminate between early and late fetal compromise, the study population was divided into three subgroups according to the gestational age at delivery (<34th gestational week; from 34th to 37th gestational weeks; and >37th gestational week).

##### 2.1. Signal Acquisition

The antepartum cCTG monitoring was performed in a controlled clinical environment with the patient lying on an armchair. The cCTG records were obtained using Corometrics 170 (General Electrics), equipped with an ultrasound transducer and a transabdominal tocodynamometer.

The Cardiotocograph was interfaced to 2CTG2 system (SEA, Italy) for computerized analysis [15] that is able to do computerized analysis on segments 3 minutes long. The FHR records were performed according to ACOG guidelines [16] and the FHR analysis was carried out using segments of 3 minutes (360 data points) without missing data, in order to prevent influences of incorrect heart rates and to obtain the same length of analysis segment for all parameters investigated, irrespective of the traces length. The initial, the middle, and the final 3 minutes of each trace were averaged, in order to obtain a single analysis segment for each trace.

The HP fetal monitors use an autocorrelation technique to compare the demodulated Doppler signal of a heartbeat with the next one. Each Doppler signal is sampled at 200 Hz (5 ms, milliseconds). The time window over which the autocorrelation function is computed is 1.2 sec, corresponding to an FHR lower bound of 50 bpm. A peak detection software then determines the heart period (the equivalent of RR period) from the autocorrelation function. With a peak position interpolation algorithm, the effective resolution is better than 2 ms.

The HP monitor produces a FHR value in bpm every 250 ms. In the commercially available system, the PC reads 10 consecutive values from the monitor every 2.5 sec and determines the actual FHR as the average of the 10 values (corresponding to an equivalent sampling frequency of 0.4 Hz). We used a modified software in order to read the FHR at 2 Hz (every 0.5 sec). The choice of reading the FHR values each 0.5 sec represents a reasonable compromise to achieve an enough large bandwidth (Nyquist frequency 1 Hz) and an acceptable accuracy of the FHR signal.

The parameters selected to quantify complexity characteristics of FHR series were the time domain parameters (Short-Term Variability, STV; Long-Term Irregularity, LTI); nonlinear parameters, such as entropy estimators (Approximate Entropy, ApEn; Sample Entropy, SampEn), Lempel Ziv Complexity (LZC); and PRSA parameters (Acceleration Phase Rectified Slope, APRS; Deceleration Phase Rectified Slope, DPRS) [17, 18].

##### 2.2. Time Domain Parameters

###### 2.2.1. Short-Term Variability

Short-Term Variability (STV) quantifies FHR variability over a very short time scale on a beat-to-beat basis [15]. Considering one minute of interbeat sequence, in ms, , we defined STV aswhere is the value of the signal taken each 2.5 sec.

###### 2.2.2. Long-Term Irregularity

Long-Term irregularity (LTI) is computed on a three-minute segment of interbeat sequence in milliseconds. Given a signal with , LTI is defined as the interquartile range (1/4; 3/4) of the distribution of the modal with :The definition is the same provided by de Haan (ACOG, 1989), with the exception of a window of 72 (and not 512) samples long. Arduini [15] excludes from the calculation big accelerations and decelerations.

##### 2.3. Nonlinear Parameters

###### 2.3.1. Entropy Estimators

Approximate Entropy (ApEn) is a collection of statistical indexes. It measures the regularity and, indirectly, the correlation and the persistence of a signal: small values indicate reduced signal irregularity. We use the original definition by Pincus (1995) [19]:where is a natural number, a positive real, and . The Approximate Entropy is computed over windows of FHR signal 3 minutes long.

Sample Entropy (SampEn) improves the estimation performed by ApEn using the same time series and parameters set. It is also the basis for a multiscale approach [20].

###### 2.3.2. Lempel Ziv Complexity

Lempel Ziv Complexity (LZC) [21] is a measure of complexity and quantifies the rate of new patterns arising with the temporal evolution over windows of FHR signal 3 minutes long. In order to estimate the LZC in a time series, it is necessary to transform the FHR signal into symbolic sequences of a finite alphabet. As a coding procedure we adopted both a binary and a ternary code. For a given time series , we construct a new sequence by mapping the original one through a binary alphabet. We symbolize with 1 a signal increase () and with 0 a decrease (). In case of ternary alphabet, 1 denotes the signal increase (), 0 the decrease (), and 2 the signal invariance (). To avoid the possible dependence of the encoded string on quantization procedure adopted to record the signal, a factor is introduced representing the minimum quantization level for a symbol change in the coded string.

##### 2.4. Phase-Rectified Signal Average (PRSA)

PRSA consists in the detection and the quantification of quasiperiodic oscillations in nonstationary signals compromised by noise and artifacts, by synchronizing the phase of all the periodic components [7]. This method can give additional information in FHR signal analysis, when episodes of increasing and/or decreasing FHR appear [22].

Acquisition and preprocessing procedure is described by Bauer and Fanelli [7, 18] and it is shown in Figure 1. The first step is the calculation of the anchor points (AP), selected according to the character that the average value of the signal before and after a certain instant within a selected time window is different. AP is valid within a time window of duration , where parameter can be used to control the upper frequency of the periodicities that are detected by PRSA. AP can be used to phase-rectify the signal, removing noise and preserving only periodic oscillations in the time series. The second step is the building of windows of 2L samples around each anchor point (L should be larger than the period of slowest oscillation that one wants to detect). In the third and fourth steps, all the 2L windows are synchronized in their anchor points and averaged, in order to obtain a single PRSA curve per patient 200 seconds long. Thus the nonperiodic components that are not synchronized with the anchor point are removed leaving only the events with a fixed phase relationship with the AP. The fifth step to identify a parameter which describes the dynamical characteristics of the curve. Bauer et al. [23] employed the Accelerations (or Decelerations) Capacity to identify a predictor for mortality after myocardial infarction. Huhn et al. [22] applied for the first time PRSA to FHR series. They employed a parameter very similar to the AC to identify and classify IUGR fetuses, called Acceleration (or Deceleration) Capacity (AAC). Fanelli et al. [18] introduced a new parameter defined as the slope of the PRSA curve computed in the AP, called Acceleration (or Deceleration) Phase-Rectified Slope (APRS or DPRS). This parameter is a descriptor of both the average increase (or decrease) in FHR amplitude (absolute change of heart frequency) and the time length of the increase (or decrease) episode.